Training system based on Bobath motion
Through the Bobath sports training system integrating intelligent devices and machine learning algorithms, the problem of traditional Bobath sports relying on high labor costs is solved, and a personalized, safe and efficient rehabilitation training plan is realized.
Patent Information
- Application Number
- CN202311357302.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-19
- Publication Date
- 2025-07-18
AI Technical Summary
Traditional Bobath sports rehabilitation training relies on a professional rehabilitation team, lacking quantitative evaluation and adjustment mechanisms, resulting in high costs and limited application.
It adopts smart gloves, angle sensors, stability stents, breathing monitors and rehabilitation software, and is connected to the central processing unit through wireless connections, integrating data analysis and machine learning algorithms to achieve real-time monitoring and automatic adjustment of rehabilitation plans.
It improves rehabilitation efficiency and quality, ensures training safety, personalizes rehabilitation plans, reduces labor costs, and expands the scope of application.
Smart Images
Figure CN120340752A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of rehabilitation medical systems, and specifically to a training system based on Bobath movement. Background Art
[0002] Rehabilitation medicine is a field that is constantly developing and innovating. Especially in the aspects of neurological rehabilitation and motor rehabilitation, traditional rehabilitation methods mainly rely on the professional knowledge and experience of physical therapists or rehabilitation doctors, and usually require a large amount of manpower and time investment. Bobath is mainly used for the rehabilitation treatment of patients with neurological injuries (such as stroke, brain injury, etc.). The implementation of traditional Bobath movement usually requires a highly professional rehabilitation team, and lacks a quantitative evaluation and adjustment mechanism, which not only increases the rehabilitation cost, but also limits its application in a wide range of occasions.
[0003] This application is based on the Bobath rehabilitation concept and modern technology, retains the advantageous features of the Bobath rehabilitation concept, and further solves the problems of difficult control of rehabilitation effect and high labor cost in the existing situation. Summary of the Invention
[0004] Technical Problems to be Solved
[0005] In view of the deficiencies of the prior art, the present invention provides a training system based on Bobath movement. The main direction of solving the problem is to solve the technical problems of the part where the Bobath handshake moves in all directions in the sitting or standing position during Bobath movement training, prevent flexion contracture of the hand, avoid wrist flexion and pronation deformity of the forearm, prevent secondary limitation of shoulder joint movement, and also help inhibit the spasm of the elbow flexor muscles.
[0006] Technical Solutions
[0007] To achieve the above object, the present invention is realized through the following technical solutions: A training system based on Bobath movement, the training system includes an intelligent glove, an angle sensor, a stability bracket, a respiratory monitor and rehabilitation software. The intelligent glove, the angle sensor, the stability bracket, the respiratory monitor and the rehabilitation software are connected to the central processing unit through a wireless connection method, and the central processing unit is built-in with data analysis and machine learning algorithms.
[0008] Preferably, the intelligent glove is equipped with pressure sensors and displacement sensors, which are respectively located in the finger, palm and wrist parts, and are used to capture hand movements and forces, and are transmitted to the central processing unit in real time through wireless connection.
[0009] Preferably, the angle sensor adopts a combination of a six-axis gyroscope and an accelerometer, with an accuracy of ±0.5 degrees, and is used to measure and control the lifting angle, and can perform data interaction with the rehabilitation software.
[0010] Preferably, the stability bracket is made of any one of carbon fiber or high-strength aluminum alloy.
[0011] Preferably, the respiration monitor uses any one of a resistive or capacitive sensor, and the measurement accuracy of the resistive or capacitive sensor is ±2% of the patient's respiration frequency and depth data, and the resistive or capacitive sensor is transmitted to the central processing unit in real time through wireless connection.
[0012] Preferably, the training system electrically connects the data of each sensor and monitor through Wi-Fi and 5G networks, and the data of the training system is electrically connected to the cloud.
[0013] Preferably, the cloud data storage uses AES-256 encryption and two-factor authentication.
[0014] Preferably, the training system incorporates a machine learning algorithm with deep learning, and the algorithm is trained using 5000 - 15000 sample points and a five-layer neural network.
[0015] Preferably, the machine learning algorithm is adjusted in real time using past rehabilitation data, current sensor data, and patient feedback.
[0016] Preferably, the specific algorithm scheme of the training system is as follows:
[0017] The adaptive rehabilitation plan adjustment algorithm includes the following parts:
[0018] Input: Patient real-time sensor data S = s1, s2, …, s n ,
[0019] Patient feedback F = f1, f2, …, f m
[0020] Output: Adjusted rehabilitation plan
[0021] P′ = p1′, p′2, …, p′ k P′ = Algorithm ML (S, F) = NN(W·S + b, F)
[0022] Where, - NN is a five-layer neural network;
[0023] - W is the weight matrix;
[0024] - b is the bias vector;
[0025] The angle out-of-safe-range alarm algorithm includes the following components
[0026] - Input: Angle sensor data A
[0027] - Output: Alarm signal
[0028]
[0029] Among them, -A max and A min are the safe upper and lower limits of the angle;
[0030] - ∈ is a small positive margin number used to provide a safety buffer;
[0031] The rehabilitation advice generation algorithm includes the following parts:
[0032] - Input: Long-term rehabilitation data of the patient D = d1, d2, …, d q
[0033] - Output: Rehabilitation advice
[0034] R = r1, r2, …, r t R = Algorithm DL (D) = CNN(G·D + c)
[0035] Among them, - CNN is a convolutional neural network;
[0036] - G is a convolutional kernel matrix;
[0037] - c is a convolutional bias..
[0038] Beneficial effects
[0039] The present invention provides a training system based on Bobath movement. It has the following beneficial effects:
[0040] 1. The system of the present invention integrates a variety of sensors and advanced data analysis algorithms. It can monitor the patient's rehabilitation status in real time and automatically adjust the rehabilitation plan. This personalized rehabilitation plan can not only improve the rehabilitation efficiency but also improve the rehabilitation quality. The system can automatically adjust the angle and pressure of the stability bracket according to the real-time data of the patient's hand movements to achieve the best rehabilitation effect.
[0041] 2. The system of the present invention conducts multi-dimensional real-time monitoring through intelligent gloves and angle sensors. In particular, the angle sensor has a high precision of ±0.5 degrees and can immediately trigger the alarm mechanism when the patient's training angle exceeds the safe range, thus ensuring the safety of the patient's training.
[0042] 3. The system of the present invention incorporates machine learning algorithms with deep learning and a real-time adjustment mechanism. This not only enables the rehabilitation plan to be automatically adjusted according to the patient's real-time feedback and sensor data, but also predicts the patient's future rehabilitation needs and progress, can self-adjust according to the patient's real-time data, and can analyze the patient's long-term rehabilitation data through deep learning algorithms to provide basic technology for subsequent upgrades and improvements. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 It is a system diagram of the present invention;
[0044] Figure 2 It is a system hardware configuration diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0045] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.
[0046] Bobath handshake: Help the patient to separate the five fingers of the affected hand, and use the healthy hand to drive the affected hand to lift. At 30 degrees, 60 degrees, 90 degrees, and 120 degrees, according to the patient's condition, keep it for about 5 - 15 minutes. It is required that the patient's hand does not shake, does not hold breath or use excessive force. The Bobath handshake training method is a commonly used method in neurodevelopmental facilitation techniques and is widely used in the rehabilitation treatment of stroke patients. The specific method of Bobath is to cross the hands and hold them, with the palms facing each other, and the thumb of the hemiplegic hand is placed on the metacarpophalangeal joint of the thumb of the healthy hand.
[0047] In the early stage of hemiplegia, the muscle strength is weak and the active movement is less. It is recommended to adopt the Bobath handshake, with the elbow joint extended and the shoulder joint flexed and lifted forward to move the upper limbs bilaterally, so as to maintain the range of motion of the shoulder joint, prevent joint contracture, and the scapula can also move to prevent shoulder subluxation. Specific Embodiment 1:
[0049] As shown in the figure, a training system based on Bobath movement. The training system includes an intelligent glove, an angle sensor, a stability bracket, a respiratory monitor, and rehabilitation software. The intelligent glove, angle sensor, stability bracket, respiratory monitor, and rehabilitation software are connected to a central processing unit through a wireless connection. The central processing unit is built with data analysis and machine learning algorithms. The intelligent glove is equipped with pressure sensors and displacement sensors, which are located in the finger, palm, and wrist parts respectively, for capturing hand movements and forces, and transmitting them to the central processing unit in real time through wireless connection. The angle sensor adopts a combination of a six-axis gyroscope and an accelerometer, with an accuracy of ±0.5 degrees, for measuring and controlling the lifting angle, and can interact with the rehabilitation software for data. The stability bracket is made of either carbon fiber or high-strength aluminum alloy. The respiratory monitor adopts either a resistive or capacitive sensor, and the measurement accuracy of the resistive or capacitive sensor is ±2% of the patient's respiratory frequency and depth data, and the resistive or capacitive sensor transmits data to the central processing unit in real time through wireless connection. The training system electrically connects the data of each sensor and monitor through Wi-Fi and 5G networks, and the data of the training system is electrically connected to the cloud. The cloud data storage uses AES-256 encryption and two-factor authentication. The training system is built with a machine learning algorithm for deep learning, and the algorithm is trained using 5000 - 15000 sample points and a five-layer neural network. The machine learning algorithm uses past rehabilitation data, current sensor data, and patient feedback for real-time adjustment. The training system is built with an HL7 standard interface. The training system is built with either an Android or iOS system, and the training system is built with rehabilitation modes: passive mode, active mode, and hybrid mode, and is electrically connected to the central processor.
[0050] The intelligent glove is equipped with pressure sensors and displacement sensors, which are located in the finger, palm, and wrist parts respectively. These sensors use advanced piezoelectric materials and optoelectronic sensing technologies, and can accurately capture the movements, forces, and angles of the hand. The angle sensor adopts a combination of a six-axis gyroscope and an accelerometer, with an accuracy of ±0.5 degrees. This high-precision measurement ability is achieved through an advanced data fusion algorithm, which can automatically calibrate and eliminate errors. The stability bracket is made of carbon fiber or high-strength aluminum alloy, with adjustable length and angle. This design enables the bracket to adapt to patients with different body types and rehabilitation needs. The respiration monitor uses resistive or capacitive sensors and can accurately measure the respiration frequency and depth of patients within a range of ±2%. This high-precision measurement is achieved through advanced signal processing technologies. The rehabilitation software is connected to the central processing unit and is built with data analysis and machine learning algorithms. These algorithms can analyze sensor data in real time, automatically adjust the rehabilitation plan, and generate detailed rehabilitation reports. Multi-sensor integration: By integrating multiple sensors (pressure, displacement, angle, respiration, etc.) in one system, comprehensive monitoring of the rehabilitation process is realized, which is rare in traditional rehabilitation equipment. High-precision measurement and control: Through the use of advanced sensors and algorithms, high-precision measurement and control of rehabilitation movements and physiological parameters are achieved. Personalized rehabilitation plan: Through machine learning algorithms, the system can automatically adjust the rehabilitation plan according to the specific conditions of each patient, which greatly improves the effectiveness and efficiency of rehabilitation. Data security and privacy protection: By using AES-256 encryption and two-factor authentication, the security and privacy of data are ensured. Multi-mode rehabilitation: The system supports passive mode, active mode, and hybrid mode, and can meet the needs of patients in different rehabilitation stages. Pressure sensors and displacement sensors: Pressure sensors usually use piezoelectric materials and can convert pressure into electrical signals; displacement sensors usually use optoelectronic or magnetoelectric technologies and can measure the movement of objects in space. Six-axis gyroscope and accelerometer: The six-axis gyroscope can measure the angular velocity and acceleration of an object on three spatial axes (X, Y, Z), while the accelerometer is specifically used to measure the acceleration of an object. Carbon fiber and high-strength aluminum alloy: Carbon fiber is a high-strength, low-density material, commonly used in high-performance applications such as aviation and racing; high-strength aluminum alloy is a specially treated aluminum alloy with excellent mechanical properties. Resistive and capacitive sensors: Resistive sensors detect physical quantities (such as temperature, pressure, etc.) by measuring changes in resistance; capacitive sensors detect physical quantities by measuring changes in capacitance. AES-256 encryption and two-factor authentication: AES-256 is an advanced encryption standard used to protect data security; two-factor authentication is a security verification method that usually requires users to provide two identity credentials (such as passwords and mobile phone verification codes). Specific Embodiment Two:
[0052] As shown in the figure, for the training system based on Bobath movement, the intelligent glove algorithm logic uses the Kalman filtering algorithm to smooth the data obtained from the pressure and displacement sensors, and uses the dynamic time warping (DTW) algorithm to identify complex gestures and movements. Initialize the sensor and Kalman filter parameters, capture the raw data, apply the Kalman filter, use the DTW algorithm for gesture recognition, and send the recognition result to the central processing unit. The angle sensor algorithm logic uses the complementary filtering algorithm to fuse the data of the gyroscope and accelerometer to accurately measure the angle. Read the gyroscope and accelerometer data, apply the complementary filtering algorithm, calculate the final angle, and send the angle data to the central processing unit. The breathing monitor algorithm logic uses the fast Fourier transform (FFT) to analyze the breathing frequency, collect the raw breathing data, apply the FFT for frequency analysis, identify the breathing frequency and depth, and send the data to the central processing unit. The rehabilitation software and the central processing unit algorithm logic use the support vector machine (SVM) to classify the patient's rehabilitation status, and use deep learning algorithms (such as RNN or LSTM) to predict the rehabilitation progress. Receive the data from each sensor, apply the SVM for rehabilitation status classification, use the deep learning algorithm for rehabilitation progress prediction, adjust the rehabilitation plan according to the prediction result. The patient wears the intelligent glove and other sensors to start the rehabilitation training. The system automatically collects the data, the central processing unit receives the data, applies the above algorithms for analysis, adjusts the rehabilitation plan according to the analysis result, updates the rehabilitation report in real time, and encrypts and stores the important data in the cloud. Such a design not only ensures a high degree of accuracy and personalization, but also realizes the comprehensive monitoring and intelligent adjustment of the rehabilitation process through advanced algorithms and... Specific Embodiment Three:
[0054] In this embodiment, the underlying algorithm logic of the specific algorithm and the code framework based on this logic are further given:
[0055] The specific algorithm scheme of the training system is as follows:
[0056] The adaptive rehabilitation plan adjustment algorithm includes the following parts:
[0057] Input: Patient real-time sensor data S = s1, s2, …, s n ,
[0058] Patient feedback F = f1, f2, …, f m
[0059] Output: Adjusted rehabilitation plan
[0060] P′ = p′1, p′2, …, p′ k P′ = Algorithm ML (S, F) = NN(W·S + b, F)
[0061] Among them, -NN is a five-layer neural network;
[0062] -W is the weight matrix;
[0063] -b is the bias vector;
[0064] The angle out-of-safe-range alarm algorithm includes the following components
[0065] - Input: Angle sensor data A
[0066] - Output: Alarm signal
[0067]
[0068] Among them, -A nax and A nin are the safe upper and lower limits of the angle;
[0069] -∈ is a small positive margin number used to provide a safety buffer;
[0070] The rehabilitation advice generation algorithm includes the following parts:
[0071] - Input: Long-term rehabilitation data D of the patient = d1, d2, …, d q
[0072] - Output: Rehabilitation advice
[0073] R = r1, r2, …, r t R = Algorithm DL (D) = CNN(G·D + c)
[0074] Among them, -CNN is a convolutional neural network;
[0075] -G is the convolutional kernel matrix;
[0076] -c is the convolutional bias.
[0077] Furthermore, the core control algorithm and the content of the core control code framework are given:
[0078] Data preprocessing: Use the Kalman filter algorithm for data smoothing. Gesture recognition: Use the Dynamic Time Warping (DTW) algorithm. Angle measurement: Use the complementary filter algorithm. Respiration monitoring: Use the Fast Fourier Transform (FFT). Rehabilitation status classification and prediction: Use the Support Vector Machine (SVM) and deep learning algorithms (such as RNN or LSTM). Core control scheme data collection and transmission: The data of all sensors is first collected and sent to the central processing unit through a wireless connection. Data analysis and processing: The central processing unit uses various core algorithms for data analysis and processing. Decision-making and adjustment: According to the data analysis results, the system automatically adjusts the rehabilitation plan. Data storage and security: Important data is encrypted and stored in the cloud.
[0079]
[0080]
[0081]
[0082] Traditional rehabilitation training equipment usually only focuses on a single body part or function. However, in this system, by integrating multiple sensors (pressure, displacement, angle, respiration, etc.), we can achieve comprehensive monitoring of the rehabilitation process. This comprehensive data collection provides the possibility for more precise rehabilitation programs. Using advanced sensors and algorithms such as Kalman filtering, dynamic time warping, complementary filtering, etc., high-precision measurement and control of rehabilitation movements and physiological parameters are achieved. This high-precision measurement and control can ensure the safety and effectiveness of the rehabilitation process. Traditional rehabilitation programs are usually fixed and cannot be adjusted according to the specific conditions of patients. However, in this system, through machine learning algorithms, the system can automatically adjust the rehabilitation program according to the specific conditions of each patient, which greatly improves the rehabilitation effect and efficiency. In the modern medical field, data security and privacy protection are very important. This system uses AES-256 encryption and two-factor authentication to ensure data security and privacy. By storing data in the cloud, doctors and rehabilitation therapists can access the patient's rehabilitation data anytime and anywhere for remote monitoring and adjustment. In addition, the cloud can also perform big data analysis to provide valuable data for rehabilitation research. Compared with traditional rehabilitation equipment, this system can collect more types of data, such as hand movements, lifting angles, respiration rates, etc., which provides the possibility for more precise rehabilitation programs. Through personalized rehabilitation programs and high-precision measurement and control, this system can ensure that each patient can obtain the best rehabilitation effect. The design of this system takes into account the comfort and convenience of patients. The intelligent glove and stability bracket are both lightweight and easy to wear and use. Compared with traditional rehabilitation equipment, this system pays more attention to data security and privacy. Through advanced encryption technology and authentication mechanisms, data security is ensured. The system is designed modularly, and function modules can be added or deleted according to needs, which provides convenience for future technology upgrades and function expansions. In short, compared with existing technologies, products, and solutions, this system has been greatly improved in terms of data collection, rehabilitation effect, user experience, data security, and scalability. Specific Embodiment Four:
[0084] The training system based on Bobath movement in this embodiment mainly includes an intelligent glove, an angle sensor, a stability bracket, a respiration monitor, and rehabilitation software. All these components are connected to the central processing unit (CPU) through a wireless connection method. The central processing unit is built-in with data analysis and machine learning algorithms.
[0085] Angle control algorithm: The PID (Proportional-Integral-Derivative) control algorithm is adopted to adjust the angle of the stability bracket in real time. When the angle sensor detects that the angle exceeds the preset threshold, the system will automatically trigger an alarm.
[0086] Pressure and Displacement Control Algorithm: Adopts fuzzy logic control to adjust the pressure of the stability bracket in real time according to the pressure and displacement sensor data of the intelligent glove.
[0087] Respiration Monitoring Algorithm: Adopts time series analysis to monitor the patient's respiration rate and depth in real time.
[0088] Machine Learning Algorithm: Adopts a five-layer neural network of deep learning, and the number of training sample points is between 5000 and 15000. This algorithm is used to make real-time adjustments according to past rehabilitation data, current sensor data and patient feedback.
[0089] Data Acquisition: The data of all sensors and monitors are transmitted to the central processing unit through Wi-Fi and 5G networks.
[0090] Data Preprocessing: In the CPU, the multi-core processor and FPGA perform data cleaning and format conversion.
[0091] Real-time Analysis and Control:
[0092] The angle control algorithm and the pressure and displacement control algorithm run in parallel in the FPGA to achieve efficient real-time control.
[0093] The respiration monitoring algorithm runs in an independent processor core.
[0094] The machine learning algorithm runs in another independent processor core.
[0095] Decision-making and Execution:
[0096] If the angle exceeds the preset threshold, an alarm is triggered.
[0097] According to the output of the machine learning algorithm, the angle and pressure of the stability bracket are automatically adjusted.
[0098] Data Storage and Security:
[0099] All data is stored in the cloud through AES-256 encryption and two-factor authentication.
[0100] Cooperation Relationship and Technical Core among Algorithms
[0101] The angle control algorithm is coordinated with the pressure and displacement control algorithm through a central scheduler to ensure that the angle and pressure of the stability bracket can be optimized simultaneously.
[0102] The output of the respiration monitoring algorithm will be used as an input feature of the machine learning algorithm to improve the personalization of the rehabilitation plan.
[0103] The machine learning algorithm is the technical core of the entire system. It synthesizes the outputs of all other algorithms, as well as the patient's historical data and real-time feedback, to achieve a highly personalized rehabilitation plan.
[0104] Angle control algorithm (PID):
[0105] Proportional (P): When the angle sensor detects a deviation from the target angle, this deviation is multiplied by a proportional constant (Kp).
[0106] Integral (I): The deviation accumulated over time is multiplied by an integral constant (Ki).
[0107] Derivative (D): The rate of change of the deviation is multiplied by a derivative constant (Kd).
[0108] Finally, these three values are added together to obtain a control signal for adjusting the angle of the stability bracket.
[0109] Pressure and displacement control algorithm (fuzzy logic):
[0110] Input: Hand movement and force data.
[0111] Fuzzy sets: Define fuzzy sets such as "low", "medium", "high", etc.
[0112] Rule: If the hand movement is "fast" and the force is "high", then the pressure is adjusted to "increase".
[0113] Defuzzification: Use the centroid method to calculate a specific control signal.
[0114] Respiration monitoring algorithm (time series analysis):
[0115] Use the sliding window method to extract respiration data over a period of time.
[0116] Apply the Fourier transform to identify the respiration frequency.
[0117] Use the autoregressive model to predict future respiration patterns.
[0118] Machine learning algorithm (deep learning):
[0119] Input layer: Various sensor data and patient feedback.
[0120] Hidden layer: A five-layer neural network using the ReLU activation function.
[0121] Output layer: Adjustment suggestions for the rehabilitation plan.
[0122] Data preprocessing
[0123] Data cleaning:
[0124] Remove outliers and noise.
[0125] Calibrate sensor data.
[0126] Data conversion:
[0127] Convert all sensor data into a unified data format (JSON or XML).
[0128] Data standardization:
[0129] Perform Z-score standardization on the data to make it have zero mean and unit variance.
[0130] Feature extraction:
[0131] Extract useful features from the original data, such as the speed and acceleration of hand movements, etc.
[0132] Cooperation relationship and technical core among algorithms (continued)
[0133] The data preprocessing module is a key step before all algorithms, which ensures the quality and consistency of the data.
[0134] The angle control algorithm and the pressure and displacement control algorithm run in parallel after data preprocessing, and their outputs are used as the inputs of the machine learning algorithm.
[0135] The machine learning algorithm integrates the outputs of all other algorithms and the patient's historical data, and makes real-time adjustments to the rehabilitation plan through a deep learning model.
[0136] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a reference structure" does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.
[0137] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A training system based on Bobath movement, characterized in that, The training system includes an intelligent glove, an angle sensor, a stability bracket, a respiratory monitor, and rehabilitation software. The intelligent glove, angle sensor, stability bracket, respiratory monitor, and rehabilitation software are connected to a central processing unit via a wireless connection. The central processing unit is built with data analysis and machine learning algorithms.
2. The Bobath movement-based training system according to claim 1, wherein the intelligent glove is equipped with pressure sensors and displacement sensors located at the finger, palm, and wrist parts respectively, for capturing hand movements and forces, and transmitting them to the central processing unit in real time via a wireless connection.
3. The Bobath movement-based training system according to claim 1, wherein the angle sensor adopts a combination of a six-axis gyroscope and an accelerometer with an accuracy of ±0.5 degrees, for measuring and controlling the lifting angle, and capable of data interaction with the rehabilitation software.
4. The Bobath movement-based training system according to claim 1, wherein the stability bracket is made of either carbon fiber or high-strength aluminum alloy.
5. The Bobath movement-based training system according to claim 1, wherein the respiratory monitor adopts either a resistive or a capacitive sensor, and the measurement accuracy of the resistive or capacitive sensor is ±2% of the patient's respiratory frequency and depth data, and the resistive or capacitive sensor transmits data to the central processing unit in real time via a wireless connection.
6. The Bobath movement-based training system according to claim 1, wherein the training system electrically connects the data of each sensor and monitor via Wi-Fi and 5G networks, and the data of the training system is electrically connected to the cloud.
7. The Bobath movement-based training system according to claim 6, wherein the cloud data storage uses AES-256 encryption and two-factor authentication.
8. The Bobath movement-based training system according to any one of claims 1-7, wherein the training system is built with a machine learning algorithm of deep learning, and the algorithm is trained using 5000-15000 sample points and a five-layer neural network.
9. The Bobath movement-based training system according to claim 8, wherein the machine learning algorithm is adjusted in real time using past rehabilitation data, current sensor data, and patient feedback.
10. The Bobath movement-based training system according to any one of claims 1-9, wherein the specific algorithm scheme of the training system is as follows: The adaptive rehabilitation plan adjustment algorithm includes the following parts: Input: Patient real-time sensor data S = s1, s2, …, s n , Patient feedback F = f1, f2, …, f m Output: Adjusted rehabilitation plan P′ = p′1, p′2, …, p′ k P′ = Algorithm ML (S,F) = NN(W·S + b,F) Among them, - NN is a five-layer neural network; - W is the weight matrix; - b is the bias vector; The angle out-of-safe-range alarm algorithm includes the following components - Input: Angle sensor data A - Output: Alarm signal Among them, -A max and A min are the safe upper and lower limits of the angle; - ∈ is a small positive margin number for providing a safety buffer; The rehabilitation advice generation algorithm includes the following parts: - Input: Long-term rehabilitation data of the patient D = d1, d2, …, d q - Output: Rehabilitation advice R = r1, r2, …, r t R = Algorithm DL (D) = CNN(G·D + c) where, - CNN is a convolutional neural network; - G is the convolution kernel matrix; - c is the convolution bias..